merlian / influxdb-zabbix

Gather data from Zabbix back-end and load to InfluxDB in near real-time for enhanced performance and easier usage with Grafana.

influxdb-zabbix

Gather data from Zabbix back-end and load to InfluxDB in near real-time for enhanced performance and easier usage with Grafana.

As InfluxDB provides an excellent compression rate (in our case: 7x), this project could be used also to archive Zabbix data.

Getting Started

• InfluxDB:

  • Install InfluxDB
  • Create a database with a retention period
    

• Grafana:

  • Install Grafana
  • Using InfluxDB in Grafana

• influxdb-zabbix:

  • Install GO
  • Setup you GOPATH
  • Run go get github.com/zensqlmonitor/influxdb-zabbix
  • Edit influxdb-zabbix.conf to match your needs
    

• PostgreSQL:

  Create user:

   CREATE USER influxdb_zabbix WITH PASSWORD '***';
   GRANT USAGE ON SCHEMA public TO influxdb_zabbix;

  Grants at the database level:

   GRANT SELECT ON public.history, public.history_uint TO influxdb_zabbix;
   GRANT SELECT ON public.trends, public.trends_uint TO influxdb_zabbix;

  Create indexes:

   CREATE UNIQUE INDEX idx_history_clock_ns_itemid
   	ON public.history USING btree (clock)
   	TABLESPACE zabbixindex;
  
   CREATE UNIQUE INDEX idx_history_uint_clock_ns_itemid
   	ON public.history_uint USING btree (clock)
   	TABLESPACE zabbixindex;
  
   CREATE INDEX idx_trends_clock_itemid
   	ON public.trends USING btree (clock)
   	TABLESPACE zabbixindex;
  
   CREATE INDEX idx_trends_uint_clock_itemid
   	ON public.trends_uint USING btree (clock)
   	TABLESPACE zabbixindex;

• MariaDB / MySQL:

  Create user:

   CREATE USER 'influxdb_zabbix'@'localhost' IDENTIFIED BY '***';

  Grants at the database level:

   GRANT SELECT ON zabbix.trends TO influxdb_zabbix@localhost;
   GRANT SELECT ON zabbix.trends_uint TO influxdb_zabbix@localhost;
   GRANT SELECT ON zabbix.history TO influxdb_zabbix@localhost;
   GRANT SELECT ON zabbix.history_uint TO influxdb_zabbix@localhost;
   flush privileges;

  Create indexes:

   CREATE UNIQUE INDEX idx_history_clock_ns_itemid
   	ON history (clock) USING btree;
  
   CREATE UNIQUE INDEX idx_history_uint_clock_ns_itemid
   	ON history_uint (clock) USING btree;
  
   CREATE INDEX idx_trends_clock_itemid
   	ON trends (clock) USING btree;
  
   CREATE INDEX idx_trends_uint_clock_itemid
   	ON trends_uint (clock) USING btree;

  NB: For trends_* tables we can use pt-online-schema-change for online index create without lock, but for history_* tables we can only use create index, because primary key for these tables does not exist.

How to use GO code

• Run in background: go run influxdb-zabbix.go &
• Build in the current directory: go build influxdb-zabbix.go
• Install in $GOPATH/bin: go install influxdb-zabbix.go

Goodies

Have a look to the scripts folder

Dependencies

• Go 1.7+
• TOML parser (https://github.com/BurntSushi/toml)
• Pure Go Postgres driver for database/sql (https://github.com/lib/pq/)
• Pure Go MySQL driver for database/sql (https://github.com/go-sql-driver/mysql/)

Configuration: influxdb-zabbix.conf

• PostgreSQL and MariaDB/MySQL supported.

• Tables that can be replicated are:

  • history
  • history_uint
  • trends
  • trends_uint

• Tables like history_log, _text and _str are not replicated.

• Configurable at table-level:

  • interval: polling interval, minimum of 15 sec
  • hours per batch : number of hours/batch to extract from zabbix backend
  • output rows per batch : allow the destination load to be splitted in multiple batches

License

MIT-LICENSE. See LICENSE file provided in the repository for details